Re: [gentoo-user] spamassassin + razor -> learning
Thomas Rösner <[EMAIL PROTECTED]> writes: > Spamassassin has an autolearn hook, but I don't know if the > razor/pyzor/etc. plugins use it, or if it's only used by the bayes > engine. I'd like to know, though, so why don't you ask on a > Spamassassin ML and report back here? ^^ Autolearn only affects the Bayes engine. To report to Razor (and Pyzor, DCC and Spamcop if configured) you have to use 'spamassassin --report' (for spam) or 'spamassassin --revoke' (to report as Ham messages which Razor has marked as spam). These also update Bayes. -- gentoo-user@gentoo.org mailing list
gentoo-user@lists.gentoo.org
Steve [Gentoo] schrieb: > One of the principle uses of my Gentoo box is to aggregate all of my > disparate pop-3 accounts, filter spam and to provide access to the ham > from my desktop and laptop using Thunderbird (both, I admit, under > Windows.) > > In the most part I am happy with this setup (postfix, dovecot, procmail, > spamassasin & ruledujour, razor, pyzor, DCC - with remote ssh tunnels > for smtp and IMAP4) - this configuration successfully filters about > 99.5% of my spam without having (yet) introduced a false positive - a > fact I attribute to the advantage that my ham is extremely unlikely to > be matched in DCC/razor/pyzor. Every day, however, I do receive a > couple of spams which I don't automatically identify. These spams are > easy enough to move out of my inbox or delete by hand, but I can't help > thinking it would be better to report these spams back to > DCC/razor/pyzor hoping to eliminate similar rubbish in future. I'm > aware of the command line client razor-report, but to be honest using > that directly is a bit of a pain...I really need a strategy to report > spam which requires no more human intervention than at present is needed > to manually remove the spam from my inbox. > > I wonder if it would be possible to set-up an "internal" account "spam" > to which only local users (i.e. myself) can send mail... "bounce" all my > spam there (using the "Mail Redirect" extension?) - and somehow use > procmail to report it? Is there a neater solution? I've written a small program in C doing something similar: Feeding user-selected spam and ham to sa-learn. Of course, it could be done with a sophisticated shell-script, but I wanted to learn C ;-) The feeding task is accomplished by shell-script, which can be edited to whatever you like. Maybe you can edit it, to feed your spam to DCC/razor/pyzor. The program and script are expecting that you use maildirformat on your IMAP-server, which is default on a gentoo/postfix-installation. Program: http://www.pc-tiede.dnsalias.net/downloads/sa-teacher-0.10.tar.bz2 ebuild: http://www.pc-tiede.dnsalias.net/downloads/sa-teacher-0.10.ebuild The category is mail-filter. Comments welcome. HTH, regards Felix -- gentoo-user@gentoo.org mailing list
gentoo-user@lists.gentoo.org
One of the principle uses of my Gentoo box is to aggregate all of my disparate pop-3 accounts, filter spam and to provide access to the ham from my desktop and laptop using Thunderbird (both, I admit, under Windows.) In the most part I am happy with this setup (postfix, dovecot, procmail, spamassasin & ruledujour, razor, pyzor, DCC - with remote ssh tunnels for smtp and IMAP4) - this configuration successfully filters about 99.5% of my spam without having (yet) introduced a false positive - a fact I attribute to the advantage that my ham is extremely unlikely to be matched in DCC/razor/pyzor. Every day, however, I do receive a couple of spams which I don't automatically identify. These spams are easy enough to move out of my inbox or delete by hand, but I can't help thinking it would be better to report these spams back to DCC/razor/pyzor hoping to eliminate similar rubbish in future. I'm aware of the command line client razor-report, but to be honest using that directly is a bit of a pain...I really need a strategy to report spam which requires no more human intervention than at present is needed to manually remove the spam from my inbox. I wonder if it would be possible to set-up an "internal" account "spam" to which only local users (i.e. myself) can send mail... "bounce" all my spam there (using the "Mail Redirect" extension?) - and somehow use procmail to report it? Is there a neater solution? Steve -- gentoo-user@gentoo.org mailing list
[gentoo-user] Re: qmail spamming
On Wednesday 04 July 2007 06:02:41, "Gentoo Voyager" <[EMAIL PROTECTED]> wrote : > Dear all, > > I'm using qmail with qmail-scanner. but i'm getting lots of mails from drug > sites(drug advertiesment) to my users mail boxes. i have installed the > spammassin. but still I'm getting such spam mail. please help me how I'm > stop this spam mails. Have you enabled dcc, razor & pyzor ? You can also use rulesdujour ... All these mechanisms help a lot while fighting against spam. I have a postfix + spamassassin + dcc + razor + pyzor + rulesdujour configuration, and spams aren't a problem for me ! Maybe you could have a look at [1] for a good tutorial on how to use these mechanisms. > Thanks & regards, > Regards, Xavier Parizet [1] http://gentoo-wiki.com/HOWTO_Email:_A_Complete_Virtual_System_-_Installing_Amavisd_and_SpamAssassin -- http://www.linuxant.fr signature.asc Description: This is a digitally signed message part.
Re: [gentoo-user] spamassassin + razor -> learning
Graham Murray wrote: Thomas Rösner <[EMAIL PROTECTED]> writes: Spamassassin has an autolearn hook, but I don't know if the razor/pyzor/etc. plugins use it, or if it's only used by the bayes engine. I'd like to know, though, so why don't you ask on a Spamassassin ML and report back here? ^^ Autolearn only affects the Bayes engine. To report to Razor (and Pyzor, DCC and Spamcop if configured) you have to use 'spamassassin --report' (for spam) or 'spamassassin --revoke' (to report as Ham messages which Razor has marked as spam). These also update Bayes. Ah, that easy. For completeness, spamc has an equivalent option: -L learn type Send message to spamd for learning. The "learn type" can be either spam, ham or forget. Must have missed it when setting it all up. Thanks & Regards, T. -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] spamassassin + razor -> learning
Thomas Rösner wrote: Graham Murray wrote: Thomas Rösner <[EMAIL PROTECTED]> writes: Spamassassin has an autolearn hook, but I don't know if the razor/pyzor/etc. plugins use it, or if it's only used by the bayes engine. I'd like to know, though, so why don't you ask on a Spamassassin ML and report back here? ^^ Autolearn only affects the Bayes engine. To report to Razor (and Pyzor, DCC and Spamcop if configured) you have to use 'spamassassin --report' (for spam) or 'spamassassin --revoke' (to report as Ham messages which Razor has marked as spam). These also update Bayes. Ah, that easy. For completeness, spamc has an equivalent option: -L learn type To correct myself: -C report type Report or revoke a message to one of the configured collaborative filtering databases. Regards, T. -- gentoo-user@gentoo.org mailing list
[gentoo-user] qmail and SA problems
im currently trying to implement spamassassin into our mail server and am running into problems im hoping you can help me with. our current setup is running gentoo/qmail/vpopmail/qmail-scanner/courier-imap/clamav/tmda. i have reinstalled qmail-scanner so that it can pickup that SA is installed, and i do have setuid for clamav installed. the problem is that none of the message seem to be being scanned by SA. its not re-writing the header with a score or anything, its like im not running anything at all. i have a testing server setup that i am doing all of this on, i have even sent the known spam test email "gtube" i think its called, and it isnt caught either. the spamd processes are running, everything looks fine. i have tried rebooting and restarting the services. i think i must have a configuration setting wrong or something. i have included the relevant parts of the config files, if there is anything else you need to help me please let me know. this has been an on-going battle for a while and i would like to find a resolution to this finally. thanks for any and all help. qmail-scanner-queue.pl: my $sa_delta='0'; my $sa_quarantine='1'; my $sa_forward=''; my $sa_fwd_verbose='0'; my $sa_delete='15'; my $sa_reject='0'; my $sa_alt='1'; my $sa_debug='1'; my $sa_hdr_report='1'; my $SA_SKIP_MD='0'; my $spamc_binary='/usr/bin/spamc'; my $spamc_options=''; my $spamc_subject="***SPAM***"; # st: if fast_spamassassin mode is selected my $spamassassin_binary=''; my $spamd_socket=''; $spamc_binary.=" -U $spamd_socket" if ($spamd_socket ne ""); my ($sa_comment,$sa_level); my $sa_symbol='+'; my ($tag_score)=""; ***am i suppose to have a scanner for SA in here somewhere? or is that the perl-scanner part?*** my @scanners_installed=("clamdscan_scanner","perlscan_scanner"); my @scanners_default=("clamdscan_scanner","perlscan_scanner"); qmail-scanner-queue-version.txt: clamdscan: 0.88/1292. perlscan: 1.25st. local.cf: required_score 6 skip_rbl_checks 0 rbl_timeout 5 # default 15 secs rewrite_header Subject *SPAM* score PYZOR_CHECK 1 score RCVD_IN_BL_SPAMCOP_NET 2.0 loadplugin Mail::SpamAssassin::Plugin::SPF loadplugin Mail::SpamAssassin::Plugin::Razor2 loadplugin Mail::SpamAssassin::Plugin::Pyzor # Razor razor_config /etc/mail/spamassassin/.razor/razor-agent.conf use_razor2 1 #Pyzor use_pyzor 1 pyzor_path /usr/sbin/pyzor #pyzor_add_header 1 # optional pyzor_timeout 15 # optional, default 10 seconds ## report_safe 1 ## use_bayes 1 bayes_path /etc/mail/spamassassin/bayes bayes_file_mode 0770 bayes_auto_learn 1 bayes_min_ham_num 400 bayes_min_spam_num 400 bayes_learn_during_report 1 bayes_use_hapaxes 1 bayes_auto_learn_threshold_nonspam 1 bayes_auto_learn_threshold_spam 14.00 bayes_ignore_header X-Bogosity bayes_ignore_header X-Spam-Flag bayes_ignore_header X-Spam-Status thanks again for your help. Nick -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] spamassassin + razor -> learning
Enrico Weigelt wrote: Hi folks, I've set up spamassassin + razor on one host, which is accessed from some other host via spamc. Is it possible to learn/report-to razor via spamc, too ? Spamassassin has an autolearn hook, but I don't know if the razor/pyzor/etc. plugins use it, or if it's only used by the bayes engine. I'd like to know, though, so why don't you ask on a Spamassassin ML and report back here? ^^ Regards, T. -- gentoo-user@gentoo.org mailing list
[gentoo-user] dcc - socket(UDP): Address family not supported by protocol
I keep getting this error whenever I invoke dcc. Now the current version is: 1.3.10 but this is not supported by Portage even when added to keyworks or unmasked, (if it was masked...:wink: ) The version listed with emerge is 1.2.74. Has anybody ever managed to get this to work properly or is there any alternative? I using Pyzor, Razor and SA, but something else would help. There was even some post about enabling various ports on my firewall to allow thru udp so that dcc could update traffic which I have done, but to no avail. In /etc/rc.conf. IP6 was enabled, but I corrected it. What should I restart to reflect this change? Will this help? Jules -- gentoo-user@gentoo.org mailing list
[gentoo-user] Please help with segfaulting python
Hi, I'm using /usr/portage/profiles/selinux/2007.0/x86 on a VIA Nehemiah box. Make.conf : CFLAGS="-Os -march=i686 -pipe -mmmx -msse" CHOST="i686-pc-linux-gnu" CXXFLAGS="${CFLAGS}" MAKEOPTS="-j2" After an emerge --deep --update --ask --verbose world which included "dev-lang/python-2.4.4-r5" emerge (pyzor and some other python script) segfault. I tried to recompile python using files in my distfiles. During building process it's look like new python binary is used (setup.py) which segfault to. I tryed different CFLAGS env var, reconfigure, without effect (it s look like they are not used during compilation). Cause every things i use not involving python work i don't think it's a memory related issue. I'm lacking of ideas and gentoo experience (trying to bypass unusable emerge). Regards, m.e. -- [EMAIL PROTECTED] mailing list
Re: [gentoo-user] cyrus-sasl, courier-authlib, which one? or both?
-BEGIN PGP SIGNED MESSAGE- Hash: SHA1 Bill Roberts wrote: > By the way, I would recommend these two guides for anyone setting up an > email server for the first time. They may not be perfect, but they are much > easier to follow than most of the howto's I've run into. > > Bill Roberts Today I've found great HOWTO copy-paste style to set postfix / mysql / courier-imap / cyrus-sasl / spamassassin / razor / pyzor / dcc / amavis / clamav. I follow it now and at this time i got succeed and to end i got to setup spamassassin. http://www.howtoforge.com/virtual_postfix_mysql_quota_courier - -- Paweł Madej aka Nysander Member of QuanTeam | RLU #357047 http://wiki.quanteam.info | Gentoo Linux User http://forum-farmaceutyczne.org | GPG key: 5861680B | keyserver: http://pgp.mit.edu Kielce, Poland | UTF-8 Email Preferred -BEGIN PGP SIGNATURE- Version: GnuPG v1.4.2 (GNU/Linux) iD8DBQFDx8khgvSMglhhaAsRAlGeAKCI8pCEaB3nvLIwaBjYI4ZGKcea0ACeJ/jV XmwbjFOjKW/IcfN9Hjm/Wlg= =w6R4 -END PGP SIGNATURE- -- gentoo-user@gentoo.org mailing list
Re: [gentoo-user] SpamAssassin is letting everything through
Mike Williams wrote: > Any chance you could enlighten the rest of us to how you enabled DCC? > Does spamassassin detect the client tools presence automagically, or is it a > compile time thing, or a config change, etc? Simply emerge dcc: * mail-filter/dcc Available versions: ~1.2.28 1.2.28-r1 ~1.2.74 Installed: no Homepage:http://www.rhyolite.com/anti-spam/dcc/ Description: Distributed Checksum Clearinghouse By default, Spamassassin is set to use all filters (DCC, Razor2 and Pyzor) by default. To specifically en-/dis-able it, add 'use_dcc 1' (or 'use_dcc 0') to your user_prefs file. -- Jonathan Wright ~ mail at djnauk.co.uk ~ www.djnauk.co.uk -- 2.6.12-gentoo-r6-djnauk-b7 Intel(R) Pentium(R) 4 Mobile CPU 1.80GHz up 15:34, 4 users, load average: 0.10, 0.13, 0.23 -- "It's hard enough to be taken seriously in the struggle for gay rights without having a bunch of straight girls running around kissing each other to get the attention of the boys and videocameras." ~ M. Robin D'Antan, 2002 -- gentoo-user@gentoo.org mailing list
[gentoo-user] Spamassassin 3.1 now fails...
A few days I 'updated world' and spamassassin 3.04 was replaced by spamassassin 3.1 - after a few minutes spent migrating my configuration (to enable Razor,Pyzor and DCC) all seemed to be working well... but then, after running fine for a day or two, the spamd daemon stopped responding and started writing an error about "syswrite(7) to the system log every few seconds: -- Nov 24 00:31:12 gifu postfix/qmgr[3563]: 9ADC2137F91: removed Nov 24 00:31:12 gifu spamd[26279]: prefork: child states: BBBII Nov 24 00:31:14 gifu postfix/smtpd[24866]: disconnect from localhost[127.0.0.1] Nov 24 00:32:40 gifu spamd[26279]: prefork: syswrite(7) failed, retrying... at / usr/lib/perl5/vendor_perl/5.8.6/Mail/SpamAssassin/SpamdForkScaling.pm line 554. Nov 24 00:32:45 gifu spamd[26279]: prefork: syswrite(7) failed, retrying... at / usr/lib/perl5/vendor_perl/5.8.6/Mail/SpamAssassin/SpamdForkScaling.pm line 554. -- Subsequently mail was (as far as I can tell) delivered without tagging messages. When I restart spamasssassin (i.e. /etc/init.d/spamd restart) the problem disappears - at least temporarily This is not a one-off occurrence - but I do not know what triggers the failure. On this occasion spamassasssin failed reporting syswrite(7) last time it reported syswrite(8) but otherwise behaviour was similar. The time at which the failure occurred does not seem to be relevant as while this time it was at just past midnight, previously it failed at 15:30. Does anyone else have this problem with Spamassassin 3.1? -- gentoo-user@gentoo.org mailing list